Added pronunciation syllable count and stress structure

Also updated CMakeLists.txt such that including projects don't have to include sqlite3.

11 files changed, 330 insertions, 12 deletions

@@ -8,3 +8,4 @@ include_directories(vendor/json/src)
 add_library(verbly lib/data.cpp lib/adjective.cpp lib/noun.cpp lib/verb.cpp lib/adverb.cpp lib/token.cpp lib/word.cpp lib/frame.cpp lib/preposition.cpp lib/adjective_query.cpp lib/adverb_query.cpp lib/noun_query.cpp lib/verb_query.cpp lib/frame_query.cpp)
 set_property(TARGET verbly PROPERTY CXX_STANDARD 11)
 set_property(TARGET verbly PROPERTY CXX_STANDARD_REQUIRED ON)
+target_link_libraries(verbly ${sqlite3_LIBRARIES})
@@ -80,6 +80,8 @@ struct pronunciation_t {
   std::string phonemes;
   std::string prerhyme;
   std::string rhyme;
+  int syllables = 0;
+  std::string stress;
   
   bool operator<(const pronunciation_t& other) const
   {
@@ -609,6 +611,8 @@ int main(int argc, char** argv)
       
       pronunciation_t p;
       p.phonemes = phonemes;
+      
+      // Rhyme detection
       if (phemstrt != std::end(phoneme_set))
       {
         std::stringstream rhymer;
@@ -641,6 +645,23 @@ int main(int argc, char** argv)
         p.rhyme = "";
       }
       
+      // Syllable/stress
+      for (auto phm : phoneme_set)
+      {
+        if (isdigit(phm.back()))
+        {
+          // It's a vowel!
+          p.syllables++;
+        
+          if (phm.back() == '1')
+          {
+            p.stress.push_back('1');
+          } else {
+            p.stress.push_back('0');
+          }
+        }
+      }
+      
       pronunciations[canonical].insert(p);
     }
   }
@@ -864,9 +885,9 @@ int main(int argc, char** argv)
         {
           if (!pronunciation.rhyme.empty())
           {
-            query = "INSERT INTO verb_pronunciations (verb_id, pronunciation, prerhyme, rhyme) VALUES (?, ?, ?, ?)";
+            query = "INSERT INTO verb_pronunciations (verb_id, pronunciation, syllables, stress, prerhyme, rhyme) VALUES (?, ?, ?, ?, ?, ?)";
           } else {
-            query = "INSERT INTO verb_pronunciations (verb_id, pronunciation) VALUES (?, ?)";
+            query = "INSERT INTO verb_pronunciations (verb_id, pronunciation, syllables, stress) VALUES (?, ?, ?, ?)";
           }
           
           if (sqlite3_prepare_v2(ppdb, query.c_str(), query.length(), &ppstmt, NULL) != SQLITE_OK)
@@ -876,11 +897,13 @@ int main(int argc, char** argv)
           
           sqlite3_bind_int(ppstmt, 1, rowid);
           sqlite3_bind_text(ppstmt, 2, pronunciation.phonemes.c_str(), pronunciation.phonemes.length(), SQLITE_TRANSIENT);
+          sqlite3_bind_int(ppstmt, 3, pronunciation.syllables);
+          sqlite3_bind_text(ppstmt, 4, pronunciation.stress.c_str(), pronunciation.stress.length(), SQLITE_TRANSIENT);
           
           if (!pronunciation.rhyme.empty())
           {
-            sqlite3_bind_text(ppstmt, 3, pronunciation.prerhyme.c_str(), pronunciation.prerhyme.length(), SQLITE_TRANSIENT);
-            sqlite3_bind_text(ppstmt, 4, pronunciation.rhyme.c_str(), pronunciation.rhyme.length(), SQLITE_TRANSIENT);
+            sqlite3_bind_text(ppstmt, 5, pronunciation.prerhyme.c_str(), pronunciation.prerhyme.length(), SQLITE_TRANSIENT);
+            sqlite3_bind_text(ppstmt, 6, pronunciation.rhyme.c_str(), pronunciation.rhyme.length(), SQLITE_TRANSIENT);
           }
           
           if (sqlite3_step(ppstmt) != SQLITE_DONE)
@@ -1243,9 +1266,9 @@ int main(int argc, char** argv)
             {
               if (!pronunciation.rhyme.empty())
               {
-                query = "INSERT INTO noun_pronunciations (noun_id, pronunciation, prerhyme, rhyme) VALUES (?, ?, ?, ?)";
+                query = "INSERT INTO noun_pronunciations (noun_id, pronunciation, syllables, stress, prerhyme, rhyme) VALUES (?, ?, ?, ?, ?, ?)";
               } else {
-                query = "INSERT INTO noun_pronunciations (noun_id, pronunciation) VALUES (?, ?)";
+                query = "INSERT INTO noun_pronunciations (noun_id, pronunciation, syllables, stress) VALUES (?, ?, ?, ?)";
               }
               
               break;
@@ -1255,9 +1278,9 @@ int main(int argc, char** argv)
             {
               if (!pronunciation.rhyme.empty())
               {
-                query = "INSERT INTO adjective_pronunciations (adjective_id, pronunciation, prerhyme, rhyme) VALUES (?, ?, ?, ?)";
+                query = "INSERT INTO adjective_pronunciations (adjective_id, pronunciation, syllables, stress, prerhyme, rhyme) VALUES (?, ?, ?, ?, ?, ?)";
               } else {
-                query = "INSERT INTO adjective_pronunciations (adjective_id, pronunciation) VALUES (?, ?)";
+                query = "INSERT INTO adjective_pronunciations (adjective_id, pronunciation, syllables, stress) VALUES (?, ?, ?, ?)";
               }
               
               break;
@@ -1267,9 +1290,9 @@ int main(int argc, char** argv)
             {
               if (!pronunciation.rhyme.empty())
               {
-                query = "INSERT INTO adverb_pronunciations (adverb_id, pronunciation, prerhyme, rhyme) VALUES (?, ?, ?, ?)";
+                query = "INSERT INTO adverb_pronunciations (adverb_id, pronunciation, syllables, stress, prerhyme, rhyme) VALUES (?, ?, ?, ?, ?, ?)";
               } else {
-                query = "INSERT INTO adverb_pronunciations (adverb_id, pronunciation) VALUES (?, ?)";
+                query = "INSERT INTO adverb_pronunciations (adverb_id, pronunciation, syllables, stress) VALUES (?, ?, ?, ?)";
               }
               
               break;
@@ -1283,11 +1306,13 @@ int main(int argc, char** argv)
         
           sqlite3_bind_int(ppstmt, 1, rowid);
           sqlite3_bind_text(ppstmt, 2, pronunciation.phonemes.c_str(), pronunciation.phonemes.length(), SQLITE_TRANSIENT);
+          sqlite3_bind_int(ppstmt, 3, pronunciation.syllables);
+          sqlite3_bind_text(ppstmt, 4, pronunciation.stress.c_str(), pronunciation.stress.length(), SQLITE_TRANSIENT);
           
           if (!pronunciation.rhyme.empty())
           {
-            sqlite3_bind_text(ppstmt, 3, pronunciation.prerhyme.c_str(), pronunciation.prerhyme.length(), SQLITE_TRANSIENT);
-            sqlite3_bind_text(ppstmt, 4, pronunciation.rhyme.c_str(), pronunciation.rhyme.length(), SQLITE_TRANSIENT);
+            sqlite3_bind_text(ppstmt, 5, pronunciation.prerhyme.c_str(), pronunciation.prerhyme.length(), SQLITE_TRANSIENT);
+            sqlite3_bind_text(ppstmt, 6, pronunciation.rhyme.c_str(), pronunciation.rhyme.length(), SQLITE_TRANSIENT);
           }
         
           if (sqlite3_step(ppstmt) != SQLITE_DONE)
@@ -186,6 +186,8 @@ CREATE TABLE `noun_pronunciations` (
   `pronunciation` VARCHAR(64) NOT NULL,
   `prerhyme` VARCHAR(8),
   `rhyme` VARCHAR(64),
+  `syllables` INT NOT NULL,
+  `stress` VARCHAR(64) NOT NULL,
   FOREIGN KEY (`noun_id`) REFERENCES `nouns`(`noun_id`)
 );
 
@@ -195,6 +197,8 @@ CREATE TABLE `verb_pronunciations` (
   `pronunciation` VARCHAR(64) NOT NULL,
   `prerhyme` VARCHAR(8),
   `rhyme` VARCHAR(64),
+  `syllables` INT NOT NULL,
+  `stress` VARCHAR(64) NOT NULL,
   FOREIGN KEY (`verb_id`) REFERENCES `verbs`(`verb_id`)
 );
 
@@ -204,6 +208,8 @@ CREATE TABLE `adjective_pronunciations` (
   `pronunciation` VARCHAR(64) NOT NULL,
   `prerhyme` VARCHAR(8),
   `rhyme` VARCHAR(64),
+  `syllables` INT NOT NULL,
+  `stress` VARCHAR(64) NOT NULL,
   FOREIGN KEY (`adjective_id`) REFERENCES `adjectives`(`adjective_id`)
 );
 
@@ -213,6 +219,8 @@ CREATE TABLE `adverb_pronunciations` (
   `pronunciation` VARCHAR(64) NOT NULL,
   `prerhyme` VARCHAR(8),
   `rhyme` VARCHAR(64),
+  `syllables` INT NOT NULL,
+  `stress` VARCHAR(64) NOT NULL,
   FOREIGN KEY (`adverb_id`) REFERENCES `adverbs`(`adverb_id`)
 );
 
@@ -88,6 +88,13 @@ namespace verbly {
     return *this;
   }
   
+  adjective_query& adjective_query::with_stress(filter<std::vector<bool>> _arg)
+  {
+    _stress = _arg;
+    
+    return *this;
+  }
+  
   adjective_query& adjective_query::with_prefix(filter<std::string> _f)
   {
     _f.clean();
@@ -338,6 +345,68 @@ namespace verbly {
       case adjective::positioning::undefined: break;
     }
     
+    if (!_stress.empty())
+    {
+      std::stringstream cond;
+      if (_stress.get_notlogic())
+      {
+        cond << "adjective_id NOT IN";
+      } else {
+        cond << "adjective_id IN";
+      }
+      
+      cond << "(SELECT adjective_id FROM adjective_pronunciations WHERE ";
+      
+      std::function<std::string (filter<std::vector<bool>>, bool)> recur = [&] (filter<std::vector<bool>> f, bool notlogic) -> std::string {
+        switch (f.get_type())
+        {
+          case filter<std::vector<bool>>::type::singleton:
+          {
+            std::ostringstream _val;
+            for (auto syl : f.get_elem())
+            {
+              if (syl)
+              {
+                _val << "1";
+              } else {
+                _val << "0";
+              }
+            }
+            
+            bindings.emplace_back(_val.str());
+            
+            if (notlogic == f.get_notlogic())
+            {
+              return "stress = ?";
+            } else {
+              return "stress != ?";
+            }
+          }
+          
+          case filter<std::vector<bool>>::type::group:
+          {
+            bool truelogic = notlogic != f.get_notlogic();
+            
+            std::list<std::string> clauses;
+            std::transform(std::begin(f), std::end(f), std::back_inserter(clauses), [&] (filter<std::vector<bool>> f2) {
+              return recur(f2, truelogic);
+            });
+            
+            if (truelogic == f.get_orlogic())
+            {
+              return "(" + verbly::implode(std::begin(clauses), std::end(clauses), " AND ") + ")";
+            } else {
+              return "(" + verbly::implode(std::begin(clauses), std::end(clauses), " OR ") + ")";
+            }
+          }
+        }
+      };
+      
+      cond << recur(_stress, _stress.get_notlogic());
+      cond << ")";
+      conditions.push_back(cond.str());
+    }
+    
     if (!_with_prefix.empty())
     {
       std::function<std::string (filter<std::string>, bool)> recur = [&] (filter<std::string> f, bool notlogic) -> std::string {
@@ -17,6 +17,7 @@ namespace verbly {
       adjective_query& has_rhyming_adjective();
       adjective_query& has_rhyming_adverb();
       adjective_query& has_rhyming_verb();
+      adjective_query& with_stress(filter<std::vector<bool>> _arg);
       
       adjective_query& requires_comparative_form();
       adjective_query& requires_superlative_form();
@@ -66,6 +67,7 @@ namespace verbly {
       bool _has_rhyming_adjective = false;
       bool _has_rhyming_adverb = false;
       bool _has_rhyming_verb = false;
+      filter<std::vector<bool>> _stress;
       
       bool _requires_comparative_form = false;
       bool _requires_superlative_form = false;
@@ -102,6 +102,13 @@ namespace verbly {
     return *this;
   }
   
+  adverb_query& adverb_query::with_stress(filter<std::vector<bool>> _arg)
+  {
+    _stress = _arg;
+    
+    return *this;
+  }
+  
   adverb_query& adverb_query::with_prefix(filter<std::string> _f)
   {
     _f.clean();
@@ -263,6 +270,68 @@ namespace verbly {
       conditions.push_back("superlative IS NOT NULL");
     }
     
+    if (!_stress.empty())
+    {
+      std::stringstream cond;
+      if (_stress.get_notlogic())
+      {
+        cond << "adverb_id NOT IN";
+      } else {
+        cond << "adverb_id IN";
+      }
+      
+      cond << "(SELECT adverb_id FROM adverb_pronunciations WHERE ";
+      
+      std::function<std::string (filter<std::vector<bool>>, bool)> recur = [&] (filter<std::vector<bool>> f, bool notlogic) -> std::string {
+        switch (f.get_type())
+        {
+          case filter<std::vector<bool>>::type::singleton:
+          {
+            std::ostringstream _val;
+            for (auto syl : f.get_elem())
+            {
+              if (syl)
+              {
+                _val << "1";
+              } else {
+                _val << "0";
+              }
+            }
+            
+            bindings.emplace_back(_val.str());
+            
+            if (notlogic == f.get_notlogic())
+            {
+              return "stress = ?";
+            } else {
+              return "stress != ?";
+            }
+          }
+          
+          case filter<std::vector<bool>>::type::group:
+          {
+            bool truelogic = notlogic != f.get_notlogic();
+            
+            std::list<std::string> clauses;
+            std::transform(std::begin(f), std::end(f), std::back_inserter(clauses), [&] (filter<std::vector<bool>> f2) {
+              return recur(f2, truelogic);
+            });
+            
+            if (truelogic == f.get_orlogic())
+            {
+              return "(" + verbly::implode(std::begin(clauses), std::end(clauses), " AND ") + ")";
+            } else {
+              return "(" + verbly::implode(std::begin(clauses), std::end(clauses), " OR ") + ")";
+            }
+          }
+        }
+      };
+      
+      cond << recur(_stress, _stress.get_notlogic());
+      cond << ")";
+      conditions.push_back(cond.str());
+    }
+    
     if (!_with_prefix.empty())
     {
       std::function<std::string (filter<std::string>, bool)> recur = [&] (filter<std::string> f, bool notlogic) -> std::string {
@@ -17,6 +17,7 @@ namespace verbly {
       adverb_query& has_rhyming_adjective();
       adverb_query& has_rhyming_adverb();
       adverb_query& has_rhyming_verb();
+      adverb_query& with_stress(filter<std::vector<bool>> _arg);
       
       adverb_query& requires_comparative_form();
       adverb_query& requires_superlative_form();
@@ -53,6 +54,7 @@ namespace verbly {
       bool _has_rhyming_adjective = false;
       bool _has_rhyming_adverb = false;
       bool _has_rhyming_verb = false;
+      filter<std::vector<bool>> _stress;
       
       bool _requires_comparative_form = false;
       bool _requires_superlative_form = false;
@@ -88,6 +88,13 @@ namespace verbly {
     return *this;
   }
   
+  noun_query& noun_query::with_stress(filter<std::vector<bool>> _arg)
+  {
+    _stress = _arg;
+    
+    return *this;
+  }
+  
   noun_query& noun_query::with_singular_form(std::string _arg)
   {
     _with_singular_form.push_back(_arg);
@@ -555,6 +562,68 @@ namespace verbly {
     {
       conditions.push_back("noun_id IN (SELECT a.noun_id FROM nouns AS a INNER JOIN noun_pronunciations AS curp ON curp.noun_id = a.noun_id INNER JOIN verb_pronunciations AS rhmp ON rhmp.prerhyme != curp.prerhyme AND rhmp.rhyme = curp.rhyme)");
     }
+    
+    if (!_stress.empty())
+    {
+      std::stringstream cond;
+      if (_stress.get_notlogic())
+      {
+        cond << "noun_id NOT IN";
+      } else {
+        cond << "noun_id IN";
+      }
+      
+      cond << "(SELECT noun_id FROM noun_pronunciations WHERE ";
+      
+      std::function<std::string (filter<std::vector<bool>>, bool)> recur = [&] (filter<std::vector<bool>> f, bool notlogic) -> std::string {
+        switch (f.get_type())
+        {
+          case filter<std::vector<bool>>::type::singleton:
+          {
+            std::ostringstream _val;
+            for (auto syl : f.get_elem())
+            {
+              if (syl)
+              {
+                _val << "1";
+              } else {
+                _val << "0";
+              }
+            }
+            
+            bindings.emplace_back(_val.str());
+            
+            if (notlogic == f.get_notlogic())
+            {
+              return "stress = ?";
+            } else {
+              return "stress != ?";
+            }
+          }
+          
+          case filter<std::vector<bool>>::type::group:
+          {
+            bool truelogic = notlogic != f.get_notlogic();
+            
+            std::list<std::string> clauses;
+            std::transform(std::begin(f), std::end(f), std::back_inserter(clauses), [&] (filter<std::vector<bool>> f2) {
+              return recur(f2, truelogic);
+            });
+            
+            if (truelogic == f.get_orlogic())
+            {
+              return "(" + verbly::implode(std::begin(clauses), std::end(clauses), " AND ") + ")";
+            } else {
+              return "(" + verbly::implode(std::begin(clauses), std::end(clauses), " OR ") + ")";
+            }
+          }
+        }
+      };
+      
+      cond << recur(_stress, _stress.get_notlogic());
+      cond << ")";
+      conditions.push_back(cond.str());
+    }
 
     for (auto except : _except)
     {
@@ -17,6 +17,7 @@ namespace verbly {
       noun_query& has_rhyming_adjective();
       noun_query& has_rhyming_adverb();
       noun_query& has_rhyming_verb();
+      noun_query& with_stress(filter<std::vector<bool>> _arg);
       
       noun_query& with_singular_form(std::string _arg);
       noun_query& with_prefix(filter<std::string> _f);
@@ -100,6 +101,7 @@ namespace verbly {
       bool _has_rhyming_adjective = false;
       bool _has_rhyming_adverb = false;
       bool _has_rhyming_verb = false;
+      filter<std::vector<bool>> _stress;
       
       std::list<std::string> _with_singular_form;
       filter<std::string> _with_prefix;
@@ -88,6 +88,13 @@ namespace verbly {
     return *this;
   }
   
+  verb_query& verb_query::with_stress(filter<std::vector<bool>> _arg)
+  {
+    _stress = _arg;
+    
+    return *this;
+  }
+  
   verb_query& verb_query::has_frames()
   {
     this->_has_frames = true;
@@ -140,6 +147,68 @@ namespace verbly {
       conditions.push_back("verb_id IN (SELECT a.verb_id FROM verbs AS a INNER JOIN verb_pronunciations AS curp ON curp.noun_id = a.adverb_id INNER JOIN verb_pronunciations AS rhmp ON rhmp.prerhyme != curp.prerhyme AND rhmp.rhyme = curp.rhyme AND rhmp.verb_id != curp.verb_id)");
     }
     
+    if (!_stress.empty())
+    {
+      std::stringstream cond;
+      if (_stress.get_notlogic())
+      {
+        cond << "verb_id NOT IN";
+      } else {
+        cond << "verb_id IN";
+      }
+      
+      cond << "(SELECT verb_id FROM verb_pronunciations WHERE ";
+      
+      std::function<std::string (filter<std::vector<bool>>, bool)> recur = [&] (filter<std::vector<bool>> f, bool notlogic) -> std::string {
+        switch (f.get_type())
+        {
+          case filter<std::vector<bool>>::type::singleton:
+          {
+            std::ostringstream _val;
+            for (auto syl : f.get_elem())
+            {
+              if (syl)
+              {
+                _val << "1";
+              } else {
+                _val << "0";
+              }
+            }
+            
+            bindings.emplace_back(_val.str());
+            
+            if (notlogic == f.get_notlogic())
+            {
+              return "stress = ?";
+            } else {
+              return "stress != ?";
+            }
+          }
+          
+          case filter<std::vector<bool>>::type::group:
+          {
+            bool truelogic = notlogic != f.get_notlogic();
+            
+            std::list<std::string> clauses;
+            std::transform(std::begin(f), std::end(f), std::back_inserter(clauses), [&] (filter<std::vector<bool>> f2) {
+              return recur(f2, truelogic);
+            });
+            
+            if (truelogic == f.get_orlogic())
+            {
+              return "(" + verbly::implode(std::begin(clauses), std::end(clauses), " AND ") + ")";
+            } else {
+              return "(" + verbly::implode(std::begin(clauses), std::end(clauses), " OR ") + ")";
+            }
+          }
+        }
+      };
+      
+      cond << recur(_stress, _stress.get_notlogic());
+      cond << ")";
+      conditions.push_back(cond.str());
+    }
+    
     for (auto except : _except)
     {
       conditions.push_back("verb_id != ?");
@@ -17,6 +17,7 @@ namespace verbly {
       verb_query& has_rhyming_adjective();
       verb_query& has_rhyming_adverb();
       verb_query& has_rhyming_verb();
+      verb_query& with_stress(filter<std::vector<bool>> _arg);
       
       verb_query& has_frames();
       
@@ -36,6 +37,7 @@ namespace verbly {
       bool _has_rhyming_adjective = false;
       bool _has_rhyming_adverb = false;
       bool _has_rhyming_verb = false;
+      filter<std::vector<bool>> _stress;
   };
 
 };